rtl-augmented

RTL Bug Fix — Augmented Dataset 是一个用于文本生成任务的数据集，主要涉及 RTL、Verilog、bug 修复和 SFT。数据集规模在 1K 到 10K 之间。数据集包含 1,130 个问题，来自 80 个仓库中的 10 个，涵盖了 109 个模块和 8 种 bug 类型。增强成功率为 59.3%。数据集的覆盖范围、问题分布、增强成功率和主题覆盖情况通过图表进行了展示。需要注意的是，某些 bug 类型（如 missing_else_latch、signal_typo 和 unconnected_port）的代表性不足，且某些主题（如 amba、amba-axi、axi 等）没有增强数据。此外，71 个已发现的仓库中没有增强数据。

The RTL Bug Fix — Augmented Dataset is a dataset tailored for text generation tasks, focusing on RTL, Verilog, bug fixing, and SFT. The dataset has a scale ranging from 1K to 10K. It contains 1,130 problem instances sourced from 10 out of 80 repositories, covering 109 modules and 8 types of bugs. The augmentation success rate reaches 59.3%. The dataset's coverage, problem distribution, augmentation success rate, and topic coverage are visualized via charts. Notably, certain bug types such as missing_else_latch, signal_typo, and unconnected_port are underrepresented, while some topics including amba, amba-axi, axi, etc., lack augmented data. Additionally, 71 of the identified repositories have no augmented data available.